In single-server deployments, data is collected for all the key performance-related metrics.
The following figure illustrates a typical high-level architecture for BMC ProactiveNet deployment using a single BMC ProactiveNet Server.
High-level architecture for single-server deployment
A typical single-server deployment can be divided into the following logical tiers:
The presentation tier includes the BMC ProactiveNet Server and embedded components such as the event cell (event processor), analytical engine, Operations Console, and the impact and service views. The presentation tier is the user interface for managing the BMC ProactiveNet for administrative functions such as event operations and performance data visualization.
While working on the presentation tier, consider the following best practices:
- Manage thresholds for all trended performance metrics.
- Visualize the performance data instead of using other consoles such as BMC PATROL or TM ART.
- Manage users with LDAP and Microsoft Active Directory.
- Use only the KPI mode of operation.
- Promote non-KPI metrics to KPI when needed, instead of operating in non-KPI mode.
- Limit the number of reports used and automatically generated to reduce related performance impact on the server.
For more information about integration, see [_Service modeling recommendations_].
Data integration tier
The data integration tier is a logical tier that manages the collection of performance data from various sources to achieve end-to-end collection of key performance data. A typical single-server deployment includes at least one data integration tier. Data integration tiers can scale out by adding additional instances, as needed, to manage the data collection.
Performance data collection is managed using the BMC ProactiveNet Integration Service. The Integration Service supports data collection from BMC PATROL Agents or from other sources such as BMC Portal, Microsoft System Center Operations Manager, and BMC TM ART using appropriate adapters. For detailed information about the Integration Service, see the BMC ProactiveNet Administrator Guide.
While working on the data integration tier, consider the following best practices:
- Install the Integration Service close to the data sources.
- Do not mix the various data sources on the same Integration Service.
- If multiple Integration Services are required from a single source, deploy BMC ProactiveNet by geography, department, business, or applications.
- Limit data collection to key performance indicators and other supportive metrics only.
- Manage enabling and disabling of the performance data at the source rather than filtering out at the adapter.
- Make use of the automated workflow feature to manage BMC PATROL data collection.
- Collect data at 5-minute polling intervals by default. Avoid faster polling frequencies.
- Review the need for a lower polling interval based on impact on the server performance and database size.
- Do not collect trends for Availability metrics through the data integration tier. Use the event tier to capture state changes for the Availability type metrics. Availability metrics such as host up or down or service up or down are not suitable for performance data collection.
Event integration tier
The event integration tier manages events from event sources such as BMC PATROL Agents, event adapters, and SNMP traps by using the Integration Service. The Integration Service includes an embedded event management cell that is installed with the Integration Service. A typical single-server deployment may contain more than one event integration tier.
Events are broadly categorized into the following sources:
- Events generated by BMC PATROL: PATROL Agents can be configured to generate events on metrics. It is best practice to configure the PATROL Agent to generate non-performance, Availability-related events, binary events, and events from the log file monitoring Knowledge Module (KM). Configure the BMC ProactiveNet Server to generate performance-related events.
- Events from other sources: Events from other sources can be processed by event adapters such as SNMP adapters and log file adapters. All events from various sources are sent to processing cells for correlation, deduplication, filtering, normalization and enrichment.
While working on the event integration tier, consider the following best practices:
- Deploy cells as required, based on event loads and event sources. For more information see Performance and scalability recommendations.
- Deploy cells close to the event source.
- Filter, enrich, normalize, deduplicate and correlate as much as possible before propagating to the next level in the data flow path.
- Deploy additional lower tier event management cells to manage high event volumes.
- Limit event messages sent from the data sources to messages that require action or analysis.
- Configure PATROL Agents to send events directly to the lower tier event management cells.
- Deploy heavily-used lower tier event management cells to dedicated Integration Service nodes. For example, if you plan to process a large number of SNMP traps as events, you should set up a dedicated event management cell to process the incoming traps.